采用LS与RBF网络的过程车削力预测与对比分析  

Prediction and contrastive analysis of process turning force using LS and RBF network

作　　者：张宝 赵春雨[2] ZHANG Bao;ZHAO Chun-yu(Xinyu University,Xinyu 338004 China;Northeastern University,Shenyang 110819 China)

机构地区：[1]新余学院机电工程学院,江西新余338004 [2]东北大学机械工程与自动化学院,辽宁沈阳110819

出　　处：《新余学院学报》2020年第5期25-30,共6页Journal of Xinyu University

摘　　要：机床车削加工过程中产生切削力是一个复杂的物理现象,其大小和方向受到很多因素的共同影响,并且在加工过程中呈现波动特性。为探究其随着主轴转速、切削深度、进给速度及进给位移的变化规律,分别运用非线性最小二乘法(LS)和径向基神经(RBF)网络对过程车削力进行预测,并对两种预测方法从精度、参数可视程度和适用性等方面进行对比分析。结果表明:采用RBF网络对过程车削力进行预测时,其预测精度高于非线性LS方法,但其参数可视化程度低于非线性LS方法。RBF网络更适用于整体的四个参数过程车削力预测,而非线性LS方法更适用于对切削参数固定而仅改变进给位移的试验组预测。The cutting forces produced in the machining process of machine tool is a complex physical phenomenon,whose size and direction are influenced by many factors,and in the process,there is a wave characteristic.In order to explore the law of the change of spindle speed,cutting depth,feed speed and feed displacement,the nonlinear least square method and radial basis function neural network were used to predict the turning force in the process,and the two prediction methods were compared and analyzed from the aspects of accuracy,parameter visibility and applicability.The results show that the accuracy of RBF network is higher than that of nonlinear LS method,but the degree of parameter visualization is lower than that of nonlinear LS method.The RBF network is more suitable for the prediction of turning force in the whole four parameter process,while the nonlinear LS method is more suitable for the prediction of the test group whose cutting parameters are fixed but only the feed displacement is changed.

关 键 词：切削试验 过程车削力建模 非线性LS RBF网络 预测对比 

分 类 号：TH113[机械工程—机械设计及理论]